• WANTED: Happy members who like to discuss audio and other topics related to our interest. Desire to learn and share knowledge of science required as is 20 years of participation in forums (not all true). Come here to have fun, be ready to be teased and not take online life too seriously. We now measure and review equipment for free! Click here for details.

Manually time-aligning subwoofer(s) to mains - how to

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
#1
This topic was opened several times on the forum but each time it got complicated and IMHO failed to produce simple how-to procedure to do this.

So, prerequisite for this procedure is that you have a device capable of adjusting mains delay and applying high pass filtering to to them in addition to providing a separate channel(s) for your subwoofer(s).
Most modern AVR receivers have such possibility, although they often offer to adjust delay only in ft/meters instead of milliseconds. If that is the case you can recalculate ms to feets or meters assuming 1ms equals to 0.343 meters/ms or 1.125ft/ms.
MiniDSP 2x4 HD is also very popular and usefull device for this job.

Note: With both of these devices (AV processor and miniDSP) it is advisable you use LFE input with your sub as that will bypass XO circuitry of your sub, which is fine as we want to have that implemented in AV proc/miniDSP. If LFE input is not available you should dial sub's XO knob to max frequency and set phase knob to 0.

Ok, so this is how you do it. In ideal case your mains should be equally distanced from your LP so time aligning a sub to one main should work for another main too. However, when time aligning a sub to a main I advise you to use the closer main to measure it and another main to play timing reference signal for both sub and main.
So, let's assume we are trying to time align a sub closer to right main, so we'll measure both of them via a band limited sweep using left main to play timing reference signal. As typical XO point used for the sub is around 80Hz band limited sweep of 30-150Hz should do fine. Use sweep length of 2M to allow more time for the steady state to stabilize.

After measuring main and the sub you should apply FDW of 6 cycles to eliminate reflections that blurr the time domain measurements but you shouldn't use "Estimate IR delay". You should unwrap phase in both measurements and use +-360deg buttons to return it to it's original position if it jumps when you unwrap it.

For this eample I used my own main speaker and measurement of F12 Rythmik sub provided by @ernestcarl. Here's how it looks (main in red, sub in blue):

Phase

Phase.JPG


IR:

IR.JPG


Step:

Step.JPG


Ok, so what we can see here from overlayed IR and step response is that mains are ahead of sub for app 12ms. As REW doesn't allow you to shift for more than 10ms at a time let's do that first. Select main measurement, press control button and "Offset t=0" button, dial -10 and apply it.

Capture.JPG


As a result of delaying the main for 10ms, phase, IR and step shifted, and this is what I got:

Phase

Phase.JPG


IR:

IR2.JPG


Step:

Step2.JPG


This looks much better with all 3 graphs. Eyeballing the step graph we can conclude that main still needs to be delayed for a little less than 2ms, so let's use the slider and dial -1.8ms.

Capture2.JPG


And this is what we are now getting:

Phase:

Phase3.JPG


IR:

IR3.JPG


Step:

Step3.JPG


So, with total delay of 11.8ms it now looks as sub is perfectly aligned to the main.

What you should do now is configure low pass filter at (say) 80Hz for the sub and equivalent high pass filter for the mains. With AV processors this usualy requires setting your mains to "small". With miniDSP you can choose XO slope, I recommend using LR24 or steeper.

If you have AV processor/receiver this is what you can do and it should sound really good, once you do FR equalisation. Unfortunately I cannot show how phase, IR and step would look with full sweep with the XO and delay implemented this way as I don't have the sub.

If you have miniDSP you can go step further and use rePhase to create a simple FIR filter which will linearise the XO between sub and mains you just created. Assuming you used LR24 at 80Hz this is how it should look, with settings I expect to work with miniDSP 2x4 HD. This should further improve things in time domain around XO so you should be able to get very nice step response with final sweep.

FIR.JPG
 
Last edited:
OP
Q

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
Thread Starter #2
I received a question via private messaging regarding use of spectrogram wavelet mode to time align sub to mains. As it may be of interest to other users I will answer the question here: As you can see I didn't use spectrogram wavelet in this method because I don't consider it precise enough. But I do recommend it as a means of control when you will be doing final 20-20kHz sweep of your system once you setup XO and delay time.

One other thing of interest to the guys who will be doing this with AV processors/receivers. When you will be entering distance figures for your front mains, center and rears use the real measured distances from your LP.

For configuring the distance to the sub: use the same figure you used for the front main to which you aligned your sub to increased by the delay you got from time alignement in REW converted into meters/ft.

So, if your left main speaker is 3 meters from your LP you enter 3 meters into the AV setup. If you measured that left main needs to be delayed by 11.8ms this means you have to enter this distance to your sub: 3 (the same as for the main) + 11.8*0,343= 7,05 m

This will tell AV processor to put your sub 11.8 ahead of your mains effectively delaying the mains for 11.8 ms.

To avoid lipsync issues you may need to delay the video for those same 11.8 msec.
 
Joined
Jun 4, 2020
Messages
55
Likes
24
#3
Thanks very much for this. Could you please also make it a bit more foolproof? For example, most people do not know what is FDW and how to set it (I found it in the other thread). At the moment I am still trying to get the very first graph you posted for Phase and cannot get it to look like yours. It would be helpful to know how to produce those steps in REW -- where to click and what options to set.

Here is my Phase chart but does not have the same direction as yours so I assume it is wrong:
Phase.png
 
OP
Q

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
Thread Starter #4
To apply FDW (frequency dependent windowing) to the selected measurement you select Tools from the main menu and choose "IR Windows". Modify only values circled in red.

Capture.JPG
 
Last edited:
OP
Q

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
Thread Starter #5
Thanks very much for this. Could you please also make it a bit more foolproof? For example, most people do not know what is FDW and how to set it (I found it in the other thread). At the moment I am still trying to get the very first graph you posted for Phase and cannot get it to look like yours. It would be helpful to know how to produce those steps in REW -- where to click and what options to set.

Here is my Phase chart but does not have the same direction as yours so I assume it is wrong:
View attachment 77758
Yes, phase should always be heading down in this frequency range, so something is very wrong.

Describe in detail what setup are you using and how you measured. Post your mdat file so I can have a look at it.

If you have multiple subs you are supposed to measure them one at a time.
 
Joined
Jun 4, 2020
Messages
55
Likes
24
#6
Yes, I am using 2 subs, and their initial distance was set by the AVR's Audyssey. I have run the setup with one sub at a time, and those are the correct distances the AVR comes up with. If one does not have a MiniDSP, then post Audyssey sub distance can be changed, and it is best advised to do that as a a group -- adding or subtracting the same distance to both subs, preserving the alignment between them as set by the AVR. In the mdat file attached you will find my Subs measurement crossed over at 80 Hz and the right speaker crossed over at 80Hz. Then, I experimented with various distance settings increasing the distance of the subs by approximately 1ms or 30cm.

I found the best distance setting to be Subs+Right 1.97 3.68 which coincidentally is the actual physical distance of my subs' cones to the MLP. Audyssey had set them to 2.47 and 4.58, but the FR over the XO region looks horrible as it cannot align subs to mains, only sub1 to sub2.
 

Attachments

Last edited:
OP
Q

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
Thread Starter #7
Yes, I am using 2 subs, and their initial distance was set by the AVR's Audyssey. I have run the setup with one sub at a time, and those are the correct distances the AVR comes up with. If one does not have a MiniDSP, then post Audyssey sub distance can be changed, and it is best advised to do that as a a group -- adding or subtracting the same distance to both subs, preserving the alignment between them as set by the AVR. In the mdat file attached you will find my Subs measurement crossed over at 80 Hz and the right speaker crossed over at 80Hz. Then, I experimented with various distance settings increasing the distance of the subs by approximately 1ms or 30cm.

I found the best distance setting to be Subs+Right 1.97 3.68 which coincidentally is the actual physical distance of my subs' cones to the MLP. Audyssey had set them to 2.47 and 4.58, but the FR over the XO region looks horrible as it cannot align subs to mains, only sub1 to sub2.
You can either let Audyssey EQ to do the time alignement job for you or you will do it yourself - but you can't have it both ways. You can't perform the procedure I described when XO is already active and delay/distances to the front main and sub are set to some values that your automatic EQ (Audyssey) system thinks are best.

I have described the procedure for the manual adjustment where you enter manually actual distance to your mains. For the initial measurement XO should be set to off, front speaker are set to "Large", subs are set to "Off" and you measure main closest to the sub using other main to play timing reference signal. Than you connect your sub to that same output channel (where front main is) and measure it. You should also ensure that Audyssey in no way interferes with your measurements.

P.S. there is nothing wrong with your phase measurements but they cannot be used to time align your sub(s) because the initial AV settings wer not as I described.
 
Joined
Jun 4, 2020
Messages
55
Likes
24
#8
I see, but then if you don't have a miniDSP you want to leverage the AVR to do EQ, which means you want to run Audyssey. Doing EQ also changes the way speakers interact. And you are still challenged then to find the right distance so you have a nice smooth transition over the crossover region. As I already mentioned, Audyssey does align the subs with each other, but not with the mains which is why you need to do the sub distance tweak to get a nice transition to the mains.

Is there a benefit then to time align subs with mains while AVR Xover is disengaged? Once bass management, Xover, and EQ are applied they will be out of phase, wouldn't they?
 
Last edited:
OP
Q

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
Thread Starter #9
From what I can see from your measurements your subs are not aligned with your mains. As long as overall FR in the 20-100Hz range is flat that may have minimal impact on sound quality, but it may as well impact how the subs are being summed with mains. If you see a dip around XO point you will know that is the case as no EQ can compensate such dip.

Now, to your question: in the example I have shown in my opening post I was aiming to time align (phase align) mains with sub around the 80Hz region as that is where sub is most frequently crossed with mains. When you connect sub to the sub chanel and apply XO at 80Hz to it and the mains and apply FR EQ that time alignement will stay intact and sub(s) will remaing being better integrated with mains.

Now, I don't know if you can make Audissey to do only EQ of the FR without touching the phase (distances) toward speakers. I also don't know what kind of phase correction Audissey does and if it would correct phase of AVR's XO at 80Hz between sub(s) and mains.
 
OP
Q

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
Thread Starter #10
I see, but then if you don't have a miniDSP you want to leverage the AVR to do EQ, which means you want to run Audyssey. Doing EQ also changes the way speakers interact. And you are still challenged then to find the right distance so you have a nice smooth transition over the crossover region. As I already mentioned, Audyssey does align the subs with each other, but not with the mains which is why you need to do the sub distance tweak to get a nice transition to the mains.

Is there a benefit then to time align subs with mains while AVR Xover is disengaged? Once bass management, Xover, and EQ are applied they will be out of phase, wouldn't they?
I would also like to add this: if I would have AVR with multiple subs where Audissey did a fine job EQ-ing my subs and rest of my speakers I wouldn't mess with time alignement of the subs.

If I would have a stereo system with setup which can run FIR filters (like miniDSP) so I can fully time align my speakers then I would also probably like that my speakers are time aligned with my sub(s) as well.

So, in your scenario, I would skip this and concentrate more on checking if Audissey has done a proper job with EQ-ing the frequency response of my mains and subs, especially in the 20-400 (or 500) Hz range.
 
Joined
Jun 4, 2020
Messages
55
Likes
24
#11
All right, thank you for all the clarifications. How about this, can you use your strategy to align two individual subs? I am attaching an mdat file which has 3 measurements:
- Sub1 only
- Sub 2 only
- Both subs

No Audyssey is engaged, AVR Xover is set at 250Hz (the maximum) and the measurement sweep runs from 20Hz to 200Hz.
 

Attachments

OP
Q

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
Thread Starter #12
All right, thank you for all the clarifications. How about this, can you use your strategy to align two individual subs? I am attaching an mdat file which has 3 measurements:
- Sub1 only
- Sub 2 only
- Both subs

No Audyssey is engaged, AVR Xover is set at 250Hz (the maximum) and the measurement sweep runs from 20Hz to 200Hz.
Time difference between those 2 subs is 4ms, but their phase response start to diverge after 80Hz so my guess is they are either of different type or one of them ("Top" sub) is sitting in a very problematic position which is causing phase to jump after 75Hz.
Response of both subs is dominated by "Top" sub.

Here's how it looks after delaying "Top" sub for 4 ms:

Step:

Step.JPG


Phase:

Phase.JPG


You can time align any number of subs but the point is that you time align sub(s) with your front main speakers as they will be crossed-over with them. If you time align 2 subs with each other that still doesn't mean they will be time aligned with your mains and the result will be sub-optimal.

As response of both subs was taken with L as reference I would also need measurement of R main with L used as timing reference to time align both subs to R main.

P.S. you also need to ensure you either use LFE input on your subs or that you have set XO on the sub panel to the max and phae to 0.
One more thing: running sweep up to 200Hz won't help. Setting sweep limits +-30Hz around planned XO frequency gives optimal result.
 
Last edited:
Joined
Jun 4, 2020
Messages
55
Likes
24
#13
Time difference between those 2 subs is 4ms, but their phase response start to diverge after 80Hz so my guess is they are either of different type or one of them ("Top" sub) is sitting in a very problematic position which is causing phase to jump after 75Hz.
Response of both subs is dominated by "Top" sub.

You can time align any number of subs but the point is that you time align sub(s) with your front main speakers as they will be crossed-over with them. If you time align 2 subs with each other that still doesn't mean they will be time aligned with your mains and the result will be sub-optimal.

As response of both subs was taken with L as reference I would also need measurement of R main with L used as timing reference to time align both subs to R main.

P.S. you also need to ensure you either use LFE input on your subs or that you have set XO on the sub panel to the max and phae to 0.
One more thing: running sweep up to 200Hz won't help. Setting sweep limits +-30Hz around planned XO frequency gives optimal result.
Thank you QMuse! Let me add that 4 ms delay and play some more. To address your comments:
- subs are identical, positioned in opposing diagonal corners of the room
- not sure which is the problematic sub as their response is very deprived above 70 Hz or so
- I understand that time aligning subs between each other is the first step, next step is to align both of them as one "virtual sub" to the mains
- subs are connected through sub outputs on AVR and are using LFE inputs; physical cross over knob on them is set to max (not active anyway)
 
Last edited:
OP
Q

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
Thread Starter #14
Thank you QMuse! Let me add that 4 ms delay and play some more. To address your comments:
- subs are identical, positioned in opposing diagonal corners of the room
- not sure which is the problematic sub as their response is very deprived above 70 Hz or so
- I understand that time aligning subs between each other is the first step, next step is to align both of them as one "virtual sub" to the mains
- subs are connected through sub outputs on AVR and are using LFE inputs; physical cross over knob on them is set to max (not active anyway)
Hard to say which one is problematic: "top" sub has smoother FR but "alc" sub has smoother phase. They will both sum very good up to 70Hz so I would consider this as a good XO point assuming your mains allow it.

- I understand that time aligning subs between each other is the first step, next step is to align both of them as one "virtual sub" to the mains
Actually I wouldn't do it that way. I would rather align one sub at a time to your mains, so try measuring your right speaker with L acting as a timing reference (as you did with both subs). This time plz use 30-100Hz sweep if you are considering 70Hz to be XO point between subs and mains.
 
OP
Q

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
Thread Starter #15
Thank you QMuse! Let me add that 4 ms delay and play some more. To address your comments:
- subs are identical, positioned in opposing diagonal corners of the room
- not sure which is the problematic sub as their response is very deprived above 70 Hz or so
- I understand that time aligning subs between each other is the first step, next step is to align both of them as one "virtual sub" to the mains
- subs are connected through sub outputs on AVR and are using LFE inputs; physical cross over knob on them is set to max (not active anyway)
Let me also add that, judging by FR "alc" sub is pretty much useless in it's position as most of the combined response is coming from "top" sub.

"alc" sub is also responsible for creating 2 dips in combined subs response at 78 and at 123Hz due to phase mismatch between them. It doesn't seem that will be corrected with time aligning them so I would recommend considering another position for "alc" sub.
 
Joined
Jun 4, 2020
Messages
55
Likes
24
#16
Let me also add that, judging by FR "alc" sub is pretty much useless in it's position as most of the combined response is coming from "top" sub.
Yeah, I am realizing that now that you highlight it. I am working on finding a new position for it. In fact, it turns out my MLP is not in a great position in the room. I get a better sub response in my left seat... so playing with the Harman room mode calculator excel sheet, but according to it 80Hz null should not exist for my room dimensions.
 
OP
Q

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
Thread Starter #17
Yeah, I am realizing that now that you highlight it. I am working on finding a new position for it. In fact, it turns out my MLP is not in a great position in the room. I get a better sub response in my left seat... so playing with the Harman room mode calculator excel sheet, but according to it 80Hz null should not exist for my room dimensions.
As a rule of a thumb, when having more than one sub you should put only one to the corner. I wouldn't put to much faith in any room mode calculator - what you measure with REW is much more accurate than you can get with any simulator.

Can you make measurements as I proposed in my 1st post: measure your "top" sub as you did (with XO set to 250Hz) and your main front which is closer to that sub (without any XO, so set it as "large" and sub to off) using other front main as a timing reference for both of these measurements?

EDIT: when doing measurement of the sub and closer main EQ should be off but sub level should be matched to the mains. Distance (delay) settings of the sub and both mains should be equal and set to the actual distance of the mains to your MLP.

There is a good chance that when both mains will sum with sub at 80Hz when properly aligned that dip at 80Hz won't be a problem at all. After we fix that you can start looking for best position of your 2nd sub, but for now keep it disconnected as it is doing more harm than good.
 
Last edited:
OP
Q

QMuse

Major Contributor
Joined
Feb 20, 2020
Messages
3,124
Likes
2,649
Thread Starter #19
All right -- here are the measurements. I did two separate ones for the sub. One of them is where the sub is set at the exact distance with the Left, and the other one is where the sub is set to its actual physical distance.
Ok, this is a consistent and well done measurement so everything went "by the book". Delaying sub by 5.1ms gives perfectly aligned phase and step. Left main is blue, sub set to same distance as main is green is sub. I also inccluded incorrectly aligned sub (orange) so you can see how much off it is.

Phase:

Phase.JPG


Step:

Step.JPG


You should now add this delay to the distance of the sub. Assuming distance for the mains and for the sub was 2.37 meters you should now set distance to the sub 2,37-5,1*0,343=0,62 meters.

It would be a good idea to repeat the sub measurement to check if your AVR did it right.

If it did, you should set XO to 80 Hz and do 40-120Hz sweep and 20-20kHz sweep to see how phase is looking when both of them (sub and main) are working together.
 
Joined
Jun 4, 2020
Messages
55
Likes
24
#20
I am not sure how you are simulating this but it seems to be working.... Attached are additional 2 measurements I did for the sub using distance at 0.62m and 4.11m (basically 5.1ms going in each direction) in the AVR. The one at 0.62 works (I think). Also, I did 40Hz - 120Hz with Xo 80Hz and a 20Hz - 20,000 Hz measurement. See the attached mdat.
 

Attachments

Last edited:
Top Bottom